OpenCattle: Nationwide Mapping of Open Cattle Feedlots Using AI-Based Object Detection

This project aims to develop and implement a high-precision, deep learning-based methodology to automatically detect open cattle feedlots across the contiguous United States using high-resolution NAIP imagery. The initiative addresses a critical data gap in confined animal area inventory by leveraging a state-of-the-art YOLO object detection model trained on thousands of hand-annotated examples.






Open cattle feedlots, vital to the U.S. livestock industry, are highly susceptible to extreme weather events that can severely impact animal welfare and productivity. Yet, comprehensive spatial data on their distribution remains scarce. This project enhances our ability to map these facilities at scale and lays the foundation for improved climate resilience planning, resource allocation, and policy-making in agricultural systems.
 

Accurate detection of animal areas across space is crucial for early disease outbreak identification, effective containment, and targeted prevention strategies to protect both animal and public health.